There’s so much discourse about polling and it seems like there’s a poll for nearly every political issue. At the same time, polls often don’t successfully help us to predict the future, including election outcomes. What contributes to the mismatch between what we expect of them and what they actually deliver? Nate Cohn is the chief political analyst at the New York Times where he created the Times/Siena poll. Cohn points out that, among many things, polling plays a “central role in the way we understand the way campaigns ought to behave.” He joins WITHpod to discuss the complexities of polling, survey methodology, systematic biases and more.
Note: This is a rough transcript — please excuse any typos.
Nate Cohn: In our sort of democracy, polling plays not just a small role, but I think a central role in the way we understand the way campaigns ought to behave. And I don’t just mean pre-election polling. Exit polls are also a poll, they’re also flawed, and that affects the calculations that political candidates make, that politicians make about what policies they’ll support or oppose.
And so, whether these numbers are accurate has real world implications besides, you know, the people who are, you know, eating their popcorn on the sidelines, you know, enjoying political campaigning as a form of entertainment.
Chris Hayes: Hello and welcome to “Why Is This Happening?” with me, your host, Chris Hayes. Well, it’s an election year and an election year means polls, polls and more polls. I would say there is an outsized amount of discussion, coverage of online discourse about polls in election years. And here’s the reason why. Human beings want to know the future, but we can’t know the future. And so ever since we came out of the caves, we’ve looked to things to tell us the future.
The stars in the sky, the tea leaves, the entrails of goats, all different kinds of ways so that we can know what will happen. We want to know what will happen. And particularly people like myself, we’re invested in the fate of the country, want to know what’s going to happen. And the closest thing that we have to that in politics are polls. But then we get mad at them because they can’t actually tell us the future. This is actually a key thing.
It’s one of the themes of today. Polls cannot tell us the future. Sometimes they tell us things that we don’t want to hear. But also, they’re methodologically somewhat complicated, and also there’s a mismatch, I think, between what they do well and what we want from them. And I think that is sort of the source of much of this frustration, dissension, you say. So, I thought it would be a great idea to like get deep into what are polls for? Like at a deep definitional level, what do we want? What is the poll?
If you think of a poll as a tool, what is it a tool to do? What do you want to use a poll to do? How do you want to understand the world? What does it help you understand about the world that you couldn’t in the absence of the poll? And I thought it’d be great to talk to the guy who’s writing about polling is probably my favorite, Nate Cohn. He’s a “New York Times” chief political analyst. He’s also the creator mastermind of the Time/Siena poll.
I find that independent of the sort of accuracy of the polling or the methodological rigor, I find Nate’s writing about how they do polling, how they think about polling, the complexities of polling, extremely lucid and engaging. And so, it’s great to have Nate Cohn on the program.
Nate Cohn: It’s great to be here, Chris. Thank you.
Chris Hayes: Do you agree with me there is so much meta discourse about polling?
Nate Cohn: I mean, the Twitter algorithm ensures that the only thing that I see in life anymore are the Seattle Mariners and polling meta discourse. So, I can’t disagree with that.
Chris Hayes: Yeah. So, there’s so much of it and I think it often can really lose sight of some real basics. And I thought one of the things we could do is just like start at the most basic thing, which is like, let’s just start polling 101. What is a poll?
Nate Cohn: A poll is an effort to scientifically sample the population and to learn public opinion by talking to a representative sample of the target population, in this case, usually Americans or American voters.
Chris Hayes: So, let’s talk about the concept of representative sample, because it’s actually a really —
Nate Cohn: It’s debated in its own right.
Chris Hayes: It’s debated in its own right. And in some ways —
Nate Cohn: Some people would say I shouldn’t have said it.
Chris Hayes: Oh, really?
Nate Cohn: Can there be a representative sample? It’s almost like an epistemological question of sorts.
Chris Hayes: Well, that is an epistemological question. I mean, let’s give an example that’s not politics. So, if you wanted to figure out what percentage of people have COVID, right, you wouldn’t want to just say how many people in the ER. What percentage of people coming into the ER have COVID, right? Because clearly you would not that would not be a representative sample.
Also, maybe you say, well, we want to send people out to randomly test folks in apartment buildings, but then you’d have to make sure that you weren’t going into neighborhoods that are particularly hard hit by COVID, right? Maybe you would sample from maternity wards across the five boroughs in every hospital. That might give you something. But even the question of what’s representative in this thought experiment I’m giving, like what rigor, what methodology, what principles does one use to establish this key adjective of representative and representative sample?
Nate Cohn: The core adjective is to know the probability that every person in the population could be included in your sample. That is the key dividing line that separates modern social science research from previous attempts in the ‘30s and earlier to try and measure the population through less systematic ways. There are a lot of different ways that you can try and obtain this kind of, quote, “probability sample” where you know the possibility that everyone can be included in the survey and what probability there was that they would be in it.
But that is one of the key dividing lines between what some people call scientific means of representative sampling and other means. There’s a middle ground, by the way, call it systematic sampling, which doesn’t necessarily give everyone an equal chance of selection, but still tries to come up with other scientific ways, other rigorous ways to try and make sure you have a decent estimate.
Chris Hayes: So, there’s some methodological means by which you want to try to get the probability as close to everyone would have an equal probability of being in the sample.
Nate Cohn: That’s the goal.
Chris Hayes: And you can measure that in certain ways.
Nate Cohn: So, some of them are theoretical. So, let’s just suppose that everyone in America has a telephone number and suppose that I can select telephone numbers at random. If that’s so, then just in principle, that means of sampling gave everyone at the moment where I selected the sample a 0.001 whatever percent chance of being included in the survey. Other means don’t depend as much on theory, but depend on your ability to observe that certain kinds of groups are represented proportionally and we’re given an opportunity to participate.
But the core of it is usually some kind of systematic, either a random sample or at least a sample where you know the varying probability that people can be included given your methodology.
Chris Hayes: So, if you could, for instance, every American citizen and green card holder has a social security number, right? So, if you could, you know, if there were some laws passed, right, the people had to answer you, right?
Nate Cohn: Yeah.
Chris Hayes: The sort of ideal platonic ideal, right. You would randomly generational security numbers to a certain number, right, that would be statistically calculated for how big your end needed to be. And then if those people had to talk to you, that would be one way of getting like a great representative sample of Americans.
Nate Cohn: That would be great. And as we’re thinking for a moment about the major ways that people conduct surveys nowadays and how closely it does or doesn’t align with that theoretical ideal you just offered. A telephone survey, it’s okay. I mean, almost everyone has a telephone number. And if we can draw a random selection of telephone numbers, we’re pretty close to that ideal.
The internet, on the other hand, is way harder. We have no idea. We have no ability to randomly generate e-mail addresses, for instance, in the same way that we could randomly generate telephone numbers. Mail, on the other hand, despite being the oldest way that polls used to be done door-to-door or face-to-face, addresses remain a viable basis for giving everyone a shot to participate in a poll.
In fact, that’s the way the census is conducted. The census, by the way, is a poll. It’s a really, really good poll, but it is a poll nonetheless that starts with every address in the country. And depending on how close you come to that ideal, and the census is probably the closest thing in our country today that does, the hope is that you have a better and better and more representative sample.
Chris Hayes: When you say telephone, so let’s talk about phone. So, there’s a lot of confusion on this point. When you say everyone has a telephone number, you mean landline and cell phone.
Nate Cohn: Correct. There are almost no polls, by the way, anymore that are landline only. That was true in a subject of considerable debate 12 years ago, but that era has come and gone. The landline is no longer even plausible as a way to reach a representative sample of Americans. And I don’t know anyone, I really don’t know any poll that exclusively relies on landlines to try and pull it off.
Chris Hayes: If I had a nickel for every time, I saw someone online comment on a poll —
Nate Cohn: I know, right? Yeah.
Chris Hayes: They’re only going to be (inaudible). I’m like, no. I promise you —
Nate Cohn: It’s not 2010 any more. Yeah.
Chris Hayes: Yeah. Okay, so it’s not just landline. So, phone numbers still work pretty well. What are the drawbacks of phone polling?
Nate Cohn: The positive element of phone polling is that it still offers this path to quickly speak to a representative sample or at least to dial a representative sample. The challenge is getting people to participate in the surveys. When we conduct a Time/Siena poll, upwards of 80 percent of the dials do not end up speaking to a human.
Chris Hayes: Wait, let me stop you right there. Eighty percent of dials do not speak to human. If you call me right now, if the Siena/Times pollster, and I love your work, what would it show up on my cell phone as?
Nate Cohn: So, the truth is that there is an ongoing fight basically between call centers and telephone providers to try and make sure that their numbers are not marked as spam and in whatever manner the call center wants it. Sometimes they want it to say Siena College. Other times they’ll spoof an area code in your zip code. You’ve undoubtedly noticed that many of the junk calls you receive are from your area.
Chris Hayes: Yeah.
Nate Cohn: That is deliberate on the part of the people who are trying to pull this off.
Chris Hayes: In fact, that’s how I know, because I have the same cell phone number from the first cell phone, I got back in 2006 or seven in Chicago, which is a 773 number and so I know. It actually is super useful.
Nate Cohn: Exactly. I know. Me, too. As a 253-area code for Seattle, you know, it’s always an indicator that it’s not someone I’m interested in speaking to. And it varies by provider. Verizon does X and then the call centers eventually realize that this means that their number is being presented that way. And then they try and adjust it.
And there’s like a sort of an ongoing effort as well to sort of determine what works best, both from a productivity standpoint and potentially a bias standpoint. I mean, I often get readers who e-mail like, if only I knew it was a Siena College poll, then I would take it. It’s like, well, you may not be the person we want to take our poll then, you know.
Chris Hayes: Yes. Right. This is precisely the problem.
Nate Cohn: There actually is some value in it not being known, even if it reduces the response rate.
Chris Hayes: Oh, I know. Hi, Siena —
Nate Cohn: Exactly.
Chris Hayes: Hey, what’s up?
Nate Cohn: Is this Nate? Is this Nate? So, it’s a constant fight to try and pull off the best version of this that can be done. And by the way, I said 80 plus percent end in no response and then an even smaller proportion actually agreed to take the survey, of course.
Chris Hayes: Right. So, 20 percent will actually pick up. And then some percentage of those are like they say, do you want to do a poll? What percentage of those people say yes?
Nate Cohn: So, there are a couple of other filters here. One is that we ask for the person who is named on a voter registration record. I know that we haven’t talked about this yet, but when we get our list of telephone numbers, we’re getting them off of voter registration files that contain a bunch of information on our respondents, like their party registration, for instance. We want to make sure that we’re talking to that person. So, we have the right number of Democrats and Republicans. As a consequence, we ask for the person listed on the voter file.
And it’s not always the person we want to. I mean, if it’s a landline, for instance, it could be someone else. These voter files are also supplemented using telephone numbers that come from external sources. So, like we think its Chris Hayes’s number based on, you know, some, I don’t know, credit card company sold —
Chris Hayes: Right.
Nate Cohn: — some third-party vendor, a list of telephone numbers, names and addresses. And then that was matched to a voter file. And then we call that number and it’s actually not Chris Hayes. And so we don’t complete the survey at that point. There are people who begin the survey and then drop off at the end or by the end. And so, when you put it all together, though, about two percent of dials of records that we attempt yield a completed interview in the end. There are bad numbers as well, of course. You know, there are a lot of different filters here.
Chris Hayes: One of the places that non-representative bias might be introduced is the toggle of who’s willing to talk to a pollster or not, right?
Nate Cohn: Absolutely.
Chris Hayes: And this is something we should spend some time on because everything you’ve done so far is like randomized, right? It’s like randomized numbers you’re going through. And then this question of this gets us a little bit back to the ER example, right? Like if I’m going to the ER to find out what percentage of the population has COVID, there’s a selection bias of people who feel sick enough to go to the ER.
Nate Cohn: Absolutely. Yeah.
Chris Hayes: So, in the analogy here, the people who are like, yes, I want to talk to a pollster might just not be representative of other people.
Nate Cohn: And in fact, we know that they’re not. You know, we know that the people who take political surveys are not surprisingly much likelier to be highly politically engaged voters who are interested in talking about this stuff with someone on the phone. One easy way that we can measure this in our polling is based on whether someone has participated in, say, a primary election. You got to be interested in politics to follow and participate in primaries, right? You know that from the low turnout.
People who participate in primaries are up twice as likely to take a telephone political survey than those who haven’t. And while you might think that telephone is the worst case there, it appears to me to be even worse online. And there are other dimensions of non-response bias as well that are even harder to measure. I think that it’s quite likely that people who just enjoy speaking to other people and who don’t —
Chris Hayes: Extroverts.
Nate Cohn: — are likelier to do that.
Chris Hayes: I got some time, hey.
Nate Cohn: I think extroverts are likelier to take a poll than those who aren’t.
Chris Hayes: Yeah.
Nate Cohn: People with privacy concerns may be less likely to take a political poll. And we could sit here and spitball about it all day, but there’s research on many of these specific anecdotes as well.
Chris Hayes: One of the, I think you wrote this analysis, but one of the things I remember from 2020 was that when Donald Trump got COVID, the polling changed dramatically. He was always in most polls and in the polling average behind Joe Biden for all of 2020. But after he got COVID, Biden shot up to 10, 12 points leads, these huge leads. When all was said and done and the election was over. I think it was you or someone else or others who went back and looked at what happened there in an attempt to answer the question, did people’s voting preference really change? Or was there something happening in the sample?
And what the conclusion was, when you went back and you looked at the people who answered and you matched to voter files, you were getting crazy numbers of hyper political folks that a certain kind of anti-Trump person was very excited to talk to a pollster in a very non-representative way that was essentially creating a false impression of what was happening in the thermometer of public opinion.
Nate Cohn: Yeah, the word to describe this is what people like to call non-response bias or partisan non-response bias, where voters from one political party or who have a certain political view like support or opposition to Donald Trump become more or less likely to respond to political surveys. And importantly, not just that they become more or less likely to respond to political surveys, but that they become more likely to respond to political surveys than other people who look just like them demographically.
We haven’t talked about this yet, but polls try to undo all of these biases by adjusting these unrepresentative samples to match the known characteristics of the population. I just told you, for instance, that people who take telephone surveys are more likely to vote in primaries. Well, that means that when you see a Time/Siena poll, we have taken all kinds of different measures to reduce the proportion of our weighted sample to include the correct number of people who have participated in the primary before.
And so, for this bias to be a problem, it doesn’t just have to be that more Democrats take a poll than Republicans. It has to be that the Democrats taking a poll are the anti-Trump Democrats and the Republicans who are taking a poll are the Haley voters and so on, while the Trump supporters become less likely to do so. And I do think there’s very strong evidence that something like that occurred not only in the immediate aftermath of Donald Trump’s diagnosis with the coronavirus, but all the way up until Election Day.
Chris Hayes: The reason I think that is important is this question of whether you’re accurately measuring or not has enormous implications for how people understand the race later. So, there’s a huge difference between Trump was down 12 points and almost won because he closed a huge gap or Susan Collins, you know, she was pulling 15 points behind.
Nate Cohn: Did she ever lead a poll? I mean, I’m not sure she ever led a poll. Yeah.
Chris Hayes: Right. Yeah, she was down the entire time. And so, then the question becomes, what did her opponent do wrong? How did they screw it up? How did they turn the ball over in the fourth quarter to give up this lead, right, which is a whole set of questions about the campaign, as opposed to it was mismeasured the whole time and the opponent was never leading.
Nate Cohn: Exactly. And, you know, you’ve just alluded to this, but, you know, in our sort of democracy, polling plays not just a small role, but I think a central role in the way we understand the way campaigns ought to behave. And I don’t just mean pre-election polling. Exit polls are also a poll. They’re also flawed. And that affects the calculations that political candidates make, that politicians make about what policies they’ll support or oppose.
And so, whether these numbers are accurate has real world implications besides, you know, the people who are eating their popcorn on the sidelines, you know, enjoying political campaigning as a form of entertainment.
Chris Hayes: More of our conversation after this quick break.
(ADVERTISEMENT)
Chris Hayes: I want to stay on this methodological question for a little longer because you’re talking about the fact that the goal is a representative sample. We can identify places where there’s possibilities for the sample to diverge from true representation. And then we adjust for it in what you call weighted samples. And we take the raw data and we put in adjustments to try to match what we’ve gotten through our actual calls with what we know about the electorate. And I think this is the place where it does feel like as much art as science, that you’re making judgment calls fundamentally.
Nate Cohn: So, I want to be a little tough on some pollsters here. There are a lot of pollsters who say that weighting is mostly an art, not a science. And I don’t really agree with that. I mean, I think that there are better and worse and more rigorous and more evidenced ways to answer this basic question about the right way to survey the voter file, which is, as I’ve already mentioned, the way that we conduct our polls, like offers a treasure trove of different options for us to understand how our surveys differ from the actual population.
And I think that most of the pollsters who have moved into the art over science world are, you know, often making an implicit concession that their scientific model has begun to break down in some way. But I mean, as far as I’m concerned, the basic decision making about how to weight a political survey is not exactly a solved problem because there are undoubtedly limits to the solution. Like the solution isn’t perfect.
But the basic mechanics here were that you are looking for ways that your sample differs from known factual characteristics about the population and that that trait is correlated with what you care about. In this case, vote choice. If those two conditions are met, you wait on it. And the limitation is that those techniques may not always be sufficient to iron out bias. You can have the right number of Democrats, but you may not have the right number of Biden voters, for instance.
Chris Hayes: Right.
Nate Cohn: You know, a primary election is a fine example of that. But the basic thought process here, I think, is one that you can go by the book on safely if you are committed to your principles and you stay within your principles.
Chris Hayes: Right. But you’re also trying to project what the universe of voters is.
Nate Cohn: Oh, sure. So that’s a separate issue, I guess. I think of turnout and what the likely electorate looks like as being kind of a separate game from the rating, adjusting the representativeness of your sample. But you’re absolutely right that the game of turnout and who’s likely to vote is a murkier game in a whole lot of different respects. And it’s absolutely vital to political polling —
Chris Hayes: Right.
Nate Cohn: — when you’re comparing an election result of a narrow universe of actual voters to a pre-election survey of a likely electorate or those who said they would vote.
Chris Hayes: But let’s not even say that. Let’s take away the likely voter screen, because that I think we would all agree there’s judgments there.
Nate Cohn: Yeah.
Chris Hayes: But, you know, particularly in a special election in Long Island and —
Nate Cohn: Oh, sure. Yeah. I mean, who knows.
Chris Hayes: — you know, who knows, right? Who knows who’s going to vote? So, let’s just say registered voters. Let’s just. —
Nate Cohn: Yeah, sure.
Chris Hayes: –you know, that’s a universe that we know. That is an established universe. We know who the registered voters in America are. The deep philosophical challenge here to me at almost like a philosophy of science level is the fact that it’s not an immutable characteristic of the universe of the categories that we use, that those are the ones that are salient.
Nate Cohn: Absolutely.
Chris Hayes: So, this is a really important point. Like you wouldn’t be looking for do we have too many left-handed people? You wouldn’t be screening for that.
Nate Cohn: Extroverts. I mean, we’ve already given examples, right? Extroverts.
Chris Hayes: Right.
Nate Cohn: What do we do about that?
Chris Hayes: Yeah.
Nate Cohn: We conducted a study where we did a poll by mail and by phone and we paid people to take the poll by mail. Got a much higher response. And we did find evidence that the people who took the poll by mail were less likely to want a job, for instance, where they would have to talk to people. So that was sort of our way of measuring this thing.
Chris Hayes: Right.
Nate Cohn: What can we do about that?
Chris Hayes: Yeah.
Nate Cohn: We don’t know the truth of what proportion of Americans want to work in a job where they speak to someone, right? We would need to know —
Chris Hayes: Right.
Nate Cohn: — something like that in order to make the statistical adjustment. So, there’s this known issue that is not ironed out here.
Chris Hayes: Right. And it’s also the fact that when you’re waiting, right, you’re comparing along lines that have political or vote choice salience, right? But those lines are shifting and constructed.
Nate Cohn: Yeah. And 2016 is a fine example here.
Chris Hayes: Yeah.
Nate Cohn: For many years, many pollsters did not weight on education —
Chris Hayes: Yes.
Nate Cohn: — was reported that people had (ph) a college degree.
Chris Hayes: Right.
Nate Cohn: And it turns out a fairly safe choice to make for decades because there was not a huge education gap in the partisan preferences of those with or without a degree. In 2016, one emerged and the pollsters that didn’t make this adjustment systematically overestimated Hillary Clinton. And it’s worth noting, by the way, that was true for years, for years or even decades. Political campaigns, politicians, journalists were all gaming out the views of the American public while probably underestimating the number of people without a degree by a large amount and probably had consequences.
Chris Hayes: Yeah. And I think the college, non-college divide is a great example of the emerging salience of a category that is, again, when we talk about these categories, we talk about crosstabs and demographics, right.
Nate Cohn: Yeah.
Chris Hayes: There’s a false surety about cleaving the world at its joints.
Nate Cohn: Oh, yeah.
Chris Hayes: These are the categories that are politically salient —
Nate Cohn: College.
Chris Hayes: — whereas all this stuff is constructed.
Nate Cohn: Even the category we’re talking about here, like educational attainment, like obscures a tremendous amount of variation within the pool of four-year college graduates. People who graduate from second tier state schools do not overwhelmingly vote (ph) Democrat.
Chris Hayes: Right.
Nate Cohn: People who graduated from Ivy League schools may be nine to one for Biden in this election. And all of these distinctions are sort of lost and we don’t adjust for those either. And by the way, this is an example of a fairly standard demographic question, one that’s on the census, one that routinely got asked in political surveys, regardless of whether people were waiting on it.
And there are all kinds of questions that we don’t ask about that could be a more salient dimension of like non-response than in the past or vote choice than in the past, regardless of whether it’s a part of non-response.
Chris Hayes: Right.
Nate Cohn: There are also some that we know matter that we’ve stopped asking. Like people don’t ask about union household membership anymore. I’m not saying that there’s a huge non-response bias there, but suppose that there ever was, you know, we’ve lost our ability to track whether we’re even accurate along this dimension.
Chris Hayes: Right. And this is why to come back to, I don’t want to overstate the art for science, but the point that I was trying to make is that —
Nate Cohn: Yeah, for sure. No, I hear where you’re coming from now.
Chris Hayes: Right. That these conceptual categories that you’re using to sort of check yourself and then apply some statistically rigorous methods are going to be shifting and it’s possible that in like the example of education, that one sneaks up on you, that you’re missing one, that you’re in the moment of formation of some new form of identity that you’re not waiting against, is my point.
Nate Cohn: Yeah. And, you know, and it’s hard to check against this to the extent we can. I mean, if you cycle through Time/Siena polls over time, you’ll see that there’s like kind of a rotating cast of demographic questions we’re kind of always asking and checking just to see whether it’s a relevant dimension for non-response compared to some sort of authoritative source that exists or at least a dimension of like partisanship that we’re not seeing. There’s a difference between a Democrat who owns a home versus rents a home, to take an example.
And that is no guarantee that we are able to identify those categories then execute a solution that ensures that the survey is representative along this dimension, at least on its face. But that’s the sort of thing that we have to be constantly trying to do.
Chris Hayes: So now I want to talk about what polling is good for. So, there is a bunch of challenges. You know, the basic idea of a representative sample is fairly clear. The hurdles that one has to jump to get from the actual people you talk to, to something that looks representative, you know, I think is clear as well and the pitfalls that might happen there. And then in the end, you produce a poll.
The poll says 38 percent of Americans think that their finances are worse off than they were five years ago, or 67 percent of Americans disapprove of Roe v. Wade being overturned. And there like really is a deep question about what that means.
Nate Cohn: Yeah.
Chris Hayes: A very deep question about what that means.
Nate Cohn: And what it’s good for. I mean, to go all the way back to the beginning, put yourself in the shoes of like the first generation of social scientists in like 1905 or whatever in the middle of the progressive era, trying to fight the dominance of big monopolies in American life and concerned about big city machines and all of these things about it. And, you know, the Senate wasn’t directly elected yet in the Constitution or whatever.
Chris Hayes: Right.
Nate Cohn: And survey research becomes this way that maybe the will of the people can become a guiding light in democracy. And over the decades to come, the basic method of reaching a random sample of Americans is refined. And the hope of these sort of progressive, I mean that in the early 20th century, meaning of the term progressive.
Chris Hayes: Yeah, right.
Nate Cohn: These progressive social scientists believe that they were offering elected officials, politicians and other Americans a window into what the people wanted and therefore, the government could represent the needs of the people better. So, abortion is like one example where, you know, the poll shows the hope of the progressive social scientists of 1912 is that a poll shows that Americans want abortion and that our politicians who are tasked with representing the people and are representative democracy look at that informs their choice and they’re likelier to enact a policy that follows the will of the people.
That’s the sort of original motive behind why do this. There are all kinds of problems that we can go into there, but that’s the sort of core first starting point. And it’s that I don’t know that it actually describes the way polling works today, to be clear. But just to start at the beginning with the first one, I think that’s the core initial animating drive behind this.
Chris Hayes: Yeah. And I think, you know, Walter Lippmann’s classic public opinion, which I’ve been referencing in every episode. I don’t know why, probably because I read it recently.
Nate Cohn: Oh, if I had known that you were referencing this stuff recently, then I didn’t know.
Chris Hayes: No, I mean, because I read it recently for the book that I’m writing. But, you know, at a deep level, what public opinion is, is a totally unsettled question and there’s not some right answer to it. It’s very complex. But your own personal opinions on things are complex. The notion there of a basic like a bat echolocating, right, that the politician or the leaders, the elected representatives are out there and the polls a way of sort of sending radar out into the world and telling them where the obstacles are, right?
Like over here, people really don’t like this, and over here, people are like, I think that’s actually super useful and a pretty important role that polling plays informationally in our democracy as a feedback mechanism.
Nate Cohn: Yeah.
Chris Hayes: And abortion right now is a great example. Like for a very long time, polling on abortion was really difficult and opaque. And that’s because the structure of people’s beliefs on it are complex and cross pressured. It depended on how you ask questions. There’s lots of people who have personal misgivings or just kind of like, eh, feel about it. But then if you ask about what the policy reaction should be, the Dobbs decision was very clarifying.
And I think a combination of actual election results where people are actually voting on a state referenda and a raft of polling now establishes, to the extent we have one, like a pretty clear view of what public opinion is on the question of —
Nate Cohn: Absolutely.
Chris Hayes: — Roe v. Wade being overturned. A majority of Americans do not like it. I think we can say that. And that’s actually a really important fact about the polity.
Nate Cohn: Yeah, absolutely. And it’s worth thinking about, like the different models for how democracy ought to work or does work and how excited you are about this, about the proposition we’ve laid here that polling can do for us. There’s a lowercase D democratic progressive era vision that is absolutely thrilled by the idea that the will of the people can sometimes be ascertained in ways that allow our government to reflect better what the electorate wants.
There’s also the more, you know, elitist lowercase R republican vision that holds that our elected officials are representing the people precisely because the people are not qualified to make policy decisions —
Chris Hayes: Correct.
Nate Cohn: — and therefore, not only should they not follow the will of the people, it may actually be bad if they did. That’s why they might oppose direct democracy, for instance. And they would, you know, then you have politicians like George W. Bush who say that, you know, they don’t follow the polls. They trust their gut and so on.
Chris Hayes: Right.
Nate Cohn: And then there’s a third. I’m not sure that it counts as being a high-level moral theory, but I think it may describe the way our democracy works more often in practice, which is like the practical way that many political actors engage in the political system in an effort to win elections. And that has huge consequences for what policies our government supports. It has huge consequences for the messages our campaigns advocate and so on. And there, pulling plays an entirely separate role, which is trying to game out what helps you win or not.
And that affects the way that policies get enacted, regardless of whether it reflects the will of the people or not. And polling ends up playing an interesting role there as well that some people may like even less than either the principled will of the people version or the lowercase R republican version
Chris Hayes: Yeah, that’s great, because that segues exactly where I wanted to go next, which is the other use of polling, which is campaigns use it and political operatives use it to do exactly what you’re saying, right, which is to figure out, to test messages, to figure out what’s front of mind, to figure out what issues people are receptive to, what attacks on their opponent rate, you know, all kinds of ways of getting feedback for this very specific end, which is to beat the opponent, right, not in the broad sense.
And also, what I think is important to distinguish here, too, just to step back is what I find most useful, if you look at the General Social Survey or you look at consumer sentiment, like repetitive longitudinal temperature checks on certain things can be very useful data sets —
Nate Cohn: Absolutely.
Chris Hayes: — to say, oh, people, you know, people got pretty bummed out after the Great Recession like that, you know.
Nate Cohn: Yeah. They didn’t like inflation.
Chris Hayes: They didn’t like inflation, right.
Nate Cohn: Yeah.
Chris Hayes: These are really useful things, but that’s a very different category than I’m running against a guy who, you know, was an orthodontist. And I want to I want to run an ad about how orthodontists suck.
Nate Cohn: Yeah.
Chris Hayes: Right? And I want to test that message, right? And a huge amount of polling in the political industry is stuff like that.
Nate Cohn: Yeah. And, like, if you’re a skeptic of this model of campaign behavior or even just public democratic engagement for good measure, like this plays out in primary elections where maybe people decide like maybe I don’t want to support Elizabeth Warren because I think she’ll lose to Trump. And there’s polling that supports that. Take an example from recent memory involving like a Time/Siena poll.
You know, there are lots of ways that this sort of horse race type polling ends up affecting political decisions of all kinds of actors, not just campaigns. Like if there’s any good news here, though, it’s like this sort of thing was happening or happens regardless of polling. One like little anecdote I like to tell people is that Abraham Lincoln was the electability candidate in 1860. People may not know this, but he was I mean, you read the books and he’s like, oh, yeah, they think he’s the moderate who can win Pennsylvania. And it’s like, I’ve heard that before.
Chris Hayes: Yes, people make these calculations.
Nate Cohn: I mean, if that’s the calculations. Yeah.
Chris Hayes: Yeah.
Nate Cohn: And that’s the weightiest, like more, in terms of moral stakes. Has there ever been, you know, an election with clear moral stakes where you would have the case to sort of toss electability aside in favor of, you know, the candidate who —
Chris Hayes: Right.
Nate Cohn: — agreed with most on the issues? And even then, in that kind of setting with the freedom of millions on the line, people are still, you know, at a convention haggling over who’s going to do best in York County or whatever. It would have been in 1860.
And there is, I think, a perfectly good case that if these calculations are being made, that polling is really helpful in informing them. I mean, we can flash forward to today with, you know, the challenge the Democrats may have among black, Hispanic and young voters that I certainly wouldn’t have believed existed if the polling weren’t telling us that, that the Biden campaign might not know if the polls weren’t telling them that. And this is an election that has huge stakes, maybe not 1860 stakes, but very large stakes by any measure. And polling will play a central role in shaping the decision of how the Biden campaign wants to win it.
Chris Hayes: Yeah. And the Trump campaign as well. I mean, I think —
Nate Cohn: And the Trump campaign as well, though, I question whether the campaign is quite as responsive to the political polling as the Biden campaign is. But I take your point.
Chris Hayes: Yeah. I mean, the Biden campaign sort of says that they think polling’s broken and I don’t know how much that’s spin and how much they actually think that.
Nate Cohn: I think you do. It’s spin. Yeah.
Chris Hayes: You think it’s spin?
Nate Cohn: Yeah.
Chris Hayes: Yeah. Let’s talk about a few. Since you brought us the head-to-head question. So, the last use of polling is the one that I think is what drives the enormous traffic. The know the future. You are the Oracle of Delphi in our modern sense. People want to know what’s going to happen and can’t know what’s going to happen. And the gap between those, they grasp for things.
Nate Cohn: And in fact, if you take the notion we just put forward seriously, these polls that people want to be predictive may be the fuel that powers decisions that change the result in the end, right?
Chris Hayes: Yes.
Nate Cohn: So, anyone who takes this stuff seriously cannot believe that they are predictive, but that’s exactly what people care about.
Chris Hayes: That’s right. So that point that they’re not predictive, they are a snapshot of time and against —
Nate Cohn: They’re measurements.
Chris Hayes: — they’re measurements and they’re measurements with an error built in and possible, again —
Nate Cohn: Oh, yeah.
Chris Hayes: — possible blind spots that we see or don’t see.
Nate Cohn: Deeply imperfect.
Chris Hayes: Yeah. Deeply imperfect.
Nate Cohn: In an era where our elections are so close, by the way —
Chris Hayes: That’s the other thing.
Nate Cohn: — in an era there is that these minor fuzzy things that have always been around now are, I mean, that almost every election, the candidate leading in the polls can lose in a way that was not true in 1984.
Chris Hayes: We’ll be right back after we take this quick break.
(ADVERTISEMENT)
Chris Hayes: I’ve written this before in a piece I wrote for “The Atlantic” after the 2020 election, but I’ll use it again here, which is if you were in a room with 100 people, 50 of them were wearing black T-shirts and 50 of them were wearing white T-shirts and then five people switch shirts, you would barely notice that. But you would be in a room that would be a landslide victory.
Nate Cohn: Yeah.
Chris Hayes: For a candidate in American politics at a national level, a 10-point win would be enormous.
Nate Cohn: It would be the largest since 1984.
Chris Hayes: Exactly. It’d be the largest since 1984. And if you were in the room surrounded and asked to eyeball, maybe you would be able to tell that there were slightly more, right?
Nate Cohn: Maybe.
Chris Hayes: So, we’re talking about margins that are so thin, everything is so far on the edge. We’re not talking about 65-35 questions. We’re not talking like —
Nate Cohn: No.
Chris Hayes: That’s one of the things actually that’s been interesting to me about the abortion stuff, where it’s not quite 65-35, but it’s a place where the signal and the noise are just so clear, like the signal is just clear as a bell and it’s showing up.
Nate Cohn: Yeah, for all we know, these abortion polls are off by way more than any recent presidential election. It doesn’t it just happens not to be material, given the size of abortion (inaudible).
Chris Hayes: Exactly. Right. So, the fact that all this stuff is happening, the margins and you have a measurement that has its error bands combines to make this all-delicate work.
Nate Cohn: Yeah.
Chris Hayes: The big question I have right now. So, I want to talk about two things specifically about the current polling, because I think people need to hear this. I have been persuaded by this point. The gap between special election performance and polling. Someone had a viral tweet that was like all political news these days is Democrats win special election for dog catcher in Hitlerburg, Montana and Trump plus six in nine new national polls.
Nate Cohn: That’s so good, but it’s true.
Chris Hayes: And I laughed at it because, and I will say this for, you know, I’ve been in this game for a little bit. Twenty years I’ve been covering politics. Before the last two years it was just the case that the party in power in the White House created a counter mobilization among the party out of power –-
Nate Cohn: Yep.
Chris Hayes: — which would show up in these special election after special election where they would get their butt kicked. And there be some headline of this state Senate district in Maryland hasn’t elected a Republican in 46 years.
Nate Cohn: Yeah.
Chris Hayes: And it has been the opposite in the Biden years to a degree that is genuinely shocking and a departure from my just my own experience in politics.
Nate Cohn: Yeah, absolutely. And not just your own experience, but this basic phenomenon we’re describing happens all over the world. If you follow like British by-elections, you know, like the liberal Dems when them all the time or whatever, you know, that’s it’s crazy. It’s not something I would have guessed was inevitable at the beginning of the Trump era.
But it’s absolutely true that based on all of the measures that we have, some of which are true facts, like the party registration of the people voting in these elections or the things that as we’ve established are not quite facts, but are nonetheless often the best things we have. The results of political surveys that the people participating in these low turnout elections are far more democratic or not just democratic. They’re far more anti-Trump —
Chris Hayes: Yeah.
Nate Cohn: — than you would guess, given the underlying sort of political fundamentals of national polling or just the reality that Democrats hold the White House.
Chris Hayes: There’s two theories for why this is, I think. One is the Dobbs effect, which is that the Dobbs decision was so seismic that it sort of turned on its head the sense of in power, out of power, basically a kind of inversion of who felt like they were winning and who felt like they were losing. And then the other is that Donald Trump is a polarizing figure along these lines that the people that are polarized against him tend to be the people that are the most, we call them high information, but the most sort of politically engaged news cycle junkies.
Nate Cohn: Yeah.
Chris Hayes: And the most like I will go to every special election. Those two factors.
Nate Cohn: There’s strong evidence for both of them, frankly. I mean, we saw after Dobbs that the special election results went from kind of middling to Democrats for two extraordinary and far better, by the way, than they fared in the midterm elections. Like the polls in the midterms were way better than the special election results where I mean, in New York, you may recall the Democrats had these fantastic performances just two months before the 2022 midterms. And then, you know, they got pummeled there a couple of months later.
So, there’s definitely evidence that Dobbs, and I would also toss into this conversation the January 6th hearings and the return of Donald Trump as another motivator on the left that has sort of reversed the typical way that things work. And the second thing you noted is also true that Donald Trump the whole time has suffered among highly engaged voters, people who regularly participate in primaries, who have a college degree and who are disproportionately white as well. Over the Trump era, Democrats have tended to gain among white voters and lose among nonwhite voters.
And there’s a third thing I’d add to this, though, and I think it’s the big twist of the election right now. Biden is doing terribly among people who aren’t regular voters.
Chris Hayes: Yes.
Nate Cohn: And that is a huge change from, say, 2012. It’s definitely not what I would have imagined in 2020 even, that not only are these highly educated voters anti-Trump, but the less engaged voters who used to be still Democratic leaning in 2020 have gone, in the polling at least. far towards Donald Trump at the moment. And obviously, that’s the sort of thing you don’t get to learn the truth of until Election Day.
But that last piece, I think, is the real thing that makes the national polls stand out so much from the specials. If less engaged voters were still voting like they did and were still backing in the polls, the same candidates they did in the 2020 election, Biden would just lead and we wouldn’t see a huge disparity.
Chris Hayes: The further you are from I vote in every election, if you move in concentric circles from like I vote in every special election to I vote in every primary to I vote in every midterm to I only vote in presidentials to I sometimes vote in presidentials.
Nate Cohn: To I’m not registered.
Chris Hayes: To I’m not registered. As you move from that center out, you drop. You move from Biden towards Trump in an incredibly dramatic fashion, according to the polling right now.
Nate Cohn: In a way that I, again, I simply wouldn’t have anticipated four years ago. Trump being weak among highly educated, highly engaged voters, that’s been true the whole time.
Chris Hayes: Yeah. Changed. Yes.
Nate Cohn: The thing that has completely upended is this the bottom falling out for Biden among the less engaged, irregular voters.
Chris Hayes: Let’s talk about that phenomenon because I’m having trouble theorizing that, and here’s why. I can very easily theorize it with essentially a monocausal theory, which is inflation. People hate inflation. Incumbent parties across the world, all of which have experienced inflation. It’s 26 percent in Hungary where Viktor Orban is. We have some of the lowest in the G7 and in the developed world, but people don’t like it and they haven’t ever experienced it.
And the people that hammers the most are people on the margins or working-class folks who are trying to keep their head above water or young people who are the most price sensitive. And so, if you think of it in that way, things didn’t used to cost this much. Now they cost this much. I liked it better when they didn’t cost that much. Donald Trump was president then. They cost too much now. Joe Biden is president now. That’s a straightforward and I think totally legible Occam’s razor theory of what’s going on, probably overly simplified, but I get that.
Here’s what I don’t get. What I don’t get is that it is clearly the case across all the data we have in the polling that between a year ago and now, Biden’s position has eroded.
Nate Cohn: Significantly.
Chris Hayes: Significantly, and among those voters, as the economic —
Nate Cohn: As inflation declines.
Chris Hayes: Yes. This is the thing. Every metric you look at, the economy has gotten tangibly better. And I totally get people being like you and you nerds with your charts. People are struggling.
Nate Cohn: Yeah.
Chris Hayes: Totally agree.
Nate Cohn: No, I —
Chris Hayes: It’s just that they were struggling more objectively a year ago.
Nate Cohn: I agree that if I were, you know, I’m imagining that there’s like a mission control for the Biden campaign with all the like maps and charts, just like NASA mission control or something. And the Biden ratings just keep dropping, even though the explanation seems, everything seems to suggest it should be going up. And I mean, I don’t know that Democrats should panic in the sense that like, you know, there’s six months, a lot of time, blah, blah, blah, blah.
All is imperfect. But like there’s something on the spacecraft that is leaking and I don’t feel like we can be quite sure we know what it is. And the economy, I think, is clearly part of it. And I think that, you know, I’m not convinced it’s the whole thing. For the reasons that you’ve mentioned, there are just too many loose ends.
I’ll go through a few of the possibilities, though, as I see it. It’s worth noting that there are a lot of overlapping demographic groups here. Young voters, less likely to vote. Also, low income, nonwhite voters, also less likely to vote, also low income. So, you have less engaged voters, young voters and nonwhite voters as these kinds of overlapping groups that each have different potential sources for weakness for Biden beyond the economy.
In the case of age, we know that Biden has never been strong among young voters. I mean, that he was did fare terribly among them in the 2020 primaries. They came around to him for the general election, an anti-Trump sentiment. But now they just straight up don’t like him. And now you have a bunch of young double haters. And right now, in the polls, these double haters are just not voting the way they say they did last time. We’ll see whether that lasts, obviously. But there has long been a disconnect here between Biden and young voters.
And so, as Biden’s ratings have collapsed, it has taken its portion of (ph) toll there. Black and Latino voters are groups that have traditionally been tied to the Democrats by identity and economic interest as much as anything. I’d argue that less ideological liberalism. In fact, many Black and Latino voters identify as moderate or conservative compared to white Democrats. We’re living through an era where a lot of the sort of racial identity politics may not be unifying across all Black and Latino voters compared to, say, the initial struggle for civil rights that brought Democrat Black voters over to Democrats as a whole.
And then finally, I’d note that this marker that we’ve been talking about of political engagement after eight years of Trump, whether you’re politically engaged is partly about what you think about Trump at this point, right? Like, someone who’s burned a lot —
Chris Hayes: Right.
Nate Cohn: — is someone who was —
Chris Hayes: That’s selecting.
Nate Cohn: — motivated to show up to stop him. And the Democratic case in this election is really boiling down to one thing at this point, which is we have to stop Donald Trump. And so, it does not surprise me that that message does not ring among the people who have not been motivated in recent elections to show up to stop him.
Chris Hayes: Well, I would slightly dissent from that, which is that I think they have been making a pretty aggressive economic message, too. It’s not as a —
Nate Cohn: A defensive one, fundamentally, though. I don’t think that when the ships are down, that the reason Democrats Joe Biden thinks he should win this election is because of the economy. I don’t I don’t think that’s their pitch here.
Chris Hayes: I disagree with that. I mean, I think they think that. I think the reason that they’re cross pressured on it is for the reason you said, which is that they’re —
Nate Cohn: Yeah.
Chris Hayes: — looking at the numbers going up and it’s not resonating. I think if you gave Joe Biden truth serum and you said, why should you be the next president, he would tell you, because I took a country that was in free fall.
Nate Cohn: Oh, I don’t agree with that. I think that if you give him the truth serum, he would say, I need to be the president because Donald Trump is a threat to the continuity of Republican government, you know. Like I think that everything that we have taken for granted as Americans like that would be the message.
Chris Hayes: Yes. Okay, both of those.
Nate Cohn: And that’s the central pitch.
Chris Hayes: They really do think they have a goodie. I mean, I know these people inside the White House economic team.
Nate Cohn: Yeah.
Chris Hayes: I mean, they’re all tearing their hair out, too, because they’re like we have overseen what feels to us like a miraculous rabbit out of the hat macroeconomic recovery that is objectively better than any pure nation like —
Nate Cohn: Yeah.
Chris Hayes: It just is true.
Nate Cohn: So, I think it’s worth positing that on paper, this is an election Biden should win for this reason, right?
Chris Hayes: Right.
Nate Cohn: He’s an incumbent president running for re-election in the economy. Even if we don’t think it’s great, like it’s not that bad. This is not a recession. And that leaves us in this position if we stipulate that to trying to figure out, okay, so why is he losing right now? And there’s one obvious explanation, which is like his age. I’m not sure that’s it, but that’s one explanation.
And the other explanation that I think would be worse for Democrats, to say the least, because it would potentially affect all Democrats, is that, you know, after the last 16 years, 12 of which with Democrats in power and the Democrats having exhausted most of their political agenda, their policy agenda.
Chris Hayes: Right.
Nate Cohn: And with issues like immigration and crime still sort of lingering out there as unresolved, perhaps in no small part because Donald Trump’s first term failed to sort of satisfy that need of the electorate, that although Biden is an incumbent president running for reelection, he’s really one running with the baggage of a very long period of Democrats either holding government or having dominated American cultural life, thanks to sort of the outsized role that Democrats had even during the Trump administration with the resistance and so on.
Chris Hayes: Right.
Nate Cohn: And that this is something more of a time for a change election that has people dissatisfied with the status quo and unlikely to ever be satisfied with the status quo, despite the fact that many of the indicators suggest they ought to be.
Chris Hayes: Yeah, I mean, I think my own theory on this, again, the problem is when I run all this against the last year, none of it works. It’s still like the leaking thing that you’re saying is like, I still can’t come. He was old a year ago. That all of the things you’ve said was true a year ago. Everything was true a year ago. So, it’s like –
Nate Cohn: One thing that has changed is that the Democrat is that the presidential race got underway.
Chris Hayes: Right.
Nate Cohn: Usually the presidential race getting underway is supposed to help the incumbent president because their party coalesces behind them and then you draw a contrast to the opponent. I don’t feel like that’s what’s been happening in this race.
Chris Hayes: Right. Right.
Nate Cohn: I think that as the presidential election got underway, Democrats started to question whether this person who’s 81 should really be running for re-election again. And that shows up in ongoing debates even now about like, should there be a contested convention or whatever?
Chris Hayes: Right.
Nate Cohn: And that debate has drawn maybe more attention to Joe Biden’s vulnerabilities than would have been focused on him in 2022. And it’s the exact opposite of what we would expect for a candidate in this state.
Chris Hayes: I will add one more thing, which you’ve written about, too, which is I think when you see an approval rating for an American polarized politics, for a politician, national politician dipping below 41, 42, they’re losing their own base.
Nate Cohn: Yeah.
Chris Hayes: And Gaza and the ongoing war there, I think, is again, we’re talking margins. But if that’s three or four points, that’s the difference between a 38 percent approval rating, 42 or 39 to 43. That’s significant.
Nate Cohn: It is.
Chris Hayes: I think there’s a pretty clear signal in the data, I would say at this point.
Nate Cohn: Yeah.
Chris Hayes: I think just based on what we’ve seen —
Nate Cohn: In the surveys that we’ve done, whether someone is a Biden defector, so to speak, someone who says that they backed them in 2020 and no longer does so is definitely related to whether they disapprove of Biden on Gaza.
Chris Hayes: The last thing I’ll say is my other theory on this is we’ll see this. My prediction is that, and this is a little bit counterintuitive, people tend to think that like attention is like oxygen for Trump. It’s like fuel on a flame. And he grows more powerful with the more attention he has. And I actually think it’s the inverse. And I think the more attention he has, the more people are repelled.
Nate Cohn: Yeah.
Chris Hayes: And my prediction is that some of that polling will come back down as he dominates more news cycles. I might be wrong, but that’s my prediction.
Nate Cohn: Well, one thing that we didn’t mention in talking about these less engaged irregular voters is exactly this. That one possibility is that the less engaged voters have soured on Biden and they’re just different than highly engaged voters who look the same. The other possibility is that they’re not engaged and not paying attention.
Chris Hayes: Right.
Nate Cohn: And that as they tune into the stakes of the election, that they’ll start to behave more like those who are engaged. And then you would end up with a scenario that I think is very real, by the way, where somehow these special elections wound up being like sort of predictive of the final result and that they drew from a highly engaged group of voters. They didn’t prove the polls wrong. The less engaged voters really don’t like Biden right now.
Chris Hayes: Right. They just didn’t vote.
Nate Cohn: Yeah, they just didn’t vote. And but as they tune in, they may start to be the hope for, I think, the Democrats and the Biden campaign is that as they’ve tuned in —
Chris Hayes: Right.
Nate Cohn: — they’ll shift. And, you know, we interview people who take our polls and a lot of these voters are not paying very close attention. We talked to someone who said abortion was her most important issue and she was voting for Trump because Roe versus Wade was overturned on Biden’s watch.
Chris Hayes: Yeah.
Nate Cohn: But someone they can, I think, reasonably hope to win that —
Chris Hayes: Yeah.
Nate Cohn: — by the end of the election, right.
Chris Hayes: Yes.
Nate Cohn: And that’s a young Latino voter in Las Vegas.
Chris Hayes: Right.
Nate Cohn: That’s the exact kind of voter who’s just sort of ground zero for Biden’s weaknesses right now.
Chris Hayes: Nate Cohn is the “New York Times” chief political analyst. He oversees the Times/Siena poll and does great writing on politics and polling. And it was a great pleasure to talk to you, Nate. That was awesome.
Nate Cohn: Great to talk to you, Chris. Thanks for having me.
Chris Hayes: Once again, great thanks to Nate Cohn, who’s an incredibly, incredibly sharp and thoughtful guy. You can give us your feedback on polling. We’d love to hear what you have to say. E-mail us at withpodgmail.com. Get in touch with us using the hashtag #WITHpod. You can follow us on TikTok by searching for WITHpod. You can follow me on Twitter, Threads and Blue Sky, all with the same handle, @chrislhayes.
“Why Is This Happening?” is presented by MSNBC and NBC News, produced by Doni Holloway and Brendan O’Melia, engineered by Bob Mallory and featuring music by Eddie Cooper. Aisha Turner is the executive producer of MSNBC Audio. You can see more of our work, including links to things we mentioned here by going to nbcnews.com/whyisthishappening.
“Why Is This Happening?” is presented by MSNBC and NBC News, produced by Doni Holloway and Brendan O’Melia, engineered by Bob Mallory and featuring music by Eddie Cooper. Aisha Turner is the executive producer of MSNBC Audio. You can see more of our work, including links to things we mentioned here by going to NBCNews.com/whyisthishappening?